Update to autoqa-0.4.0-1

[James Laska]

Greetings,

As previously discussed, I'd like to update the autoqa.spec to capture the
0.4.0 release changes.  Thanks to wwoods for helping create a pared down (but
accurate) %changelog.  I've tested the changes so far on
autoqa.fedoraproject.org (based on commit
5251bdccebb9a9612f19fa008276ebc5a07ae29c as suggested by Kamil).  Note,
autoqa.fp.org is deployed as an EPEL5 system, so compatibility with python-2.4
is needed.

Once Kamil has addressed the multihook patches to his liking, I can build
autoqa-0.4.1-1 and deploy.  Questions/concerns/ideas?

If no concerns, I'll commit these patches to master later this week.

Thanks,
James

[PATCH 1/2] Update to autoqa-0.4.0-1

[James Laska]

---
 autoqa.spec |   19 ++++++++++++++++++-
 1 files changed, 18 insertions(+), 1 deletions(-)

diff --git a/autoqa.spec b/autoqa.spec
index 8e60be8..c75bcad 100644
--- a/autoqa.spec
+++ b/autoqa.spec
@@ -5,7 +5,7 @@

Name: autoqa
 Summary: Automated quality assurance framework
-Version: 0.3.5
+Version: 0.4.0
 Release: 1%{?dist}
 Source0: http://fedorahosted.org/autoqa/releases/%%7Bname%7D-%%7Bversion%7D.tar.gz
 License: GPLv2+
@@ -100,6 +100,23 @@ rm -rf $RPM_BUILD_ROOT

%changelog
+* Tue Aug 30 2010 James Laska  - 0.4.0-1
+- Ticket#123 - Add upgrade-path test (vaschenb)
+- Ticket#195 - Add Helloworld test for hook verification (vaschenb)
+- Ticket#199 - Standardize test arguments between hooks (kparal)
+- Ticket#200 - Support for control.autoqa (kparal)
+- Ticket#207 - Use correct list of tags when looking for koji builds (jlaska)
+- Ticket#210 - renamed url parameter to baseurl (jskladan)
+- Ticket#211 - update repoinfo.conf for new F-14 branch repo (jlaska)
+- Replace repoinfo.conf on upgrade (kparal)
+- New AutoQATest base class for test objects (jskladan)
+- Add post-bodhi-update watcher to crontab (kparal)
+- Treat ghost files appropriately in potential_conflicts.py, i.e. match rpm's
+  behaviour (fcami)
+- Add repoinfo.getreleases() and get_bodhi_releases() (wwoods)
+- Use repoinfo data for bodhi_releases in the bodhi watcher (wwoods)
+- Add support for autotest labels in autoqa harness (kparal)
+
 * Tue Jul 06 2010 James Laska  - 0.3.5-1
 - Move common methods from post-bodhi-update into bodhi_utils.py (wwoods)
 - Require autotest, not autotest-client (jlaska)

[PATCH 2/2] Update autoqa and watch-bodhi-requests.py to be python-2.4 friendly so it can run on EPEL5.

[James Laska]

---
 autoqa                                          |    2 +-
 hooks/post-bodhi-update/watch-bodhi-requests.py |    4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/autoqa b/autoqa
index e0848ea..e6bd8c9 100755
--- a/autoqa
+++ b/autoqa
@@ -258,7 +258,7 @@ test_vars = {} # dict of test->its test vars
 for test in tests[:]:
     try:
         test_vars[test] = eval_test_vars(test, default_test_vars)
-    except IOError as e:
+    except IOError, e:
         print "Error: Can't evaluate test '%s': %s" % (test, e)
         tests.remove(test)
 testlist = [test for test,vars in test_vars.iteritems() if vars['execute'] == True]
diff --git a/hooks/post-bodhi-update/watch-bodhi-requests.py b/hooks/post-bodhi-update/watch-bodhi-requests.py
index 45f168f..597853a 100755
--- a/hooks/post-bodhi-update/watch-bodhi-requests.py
+++ b/hooks/post-bodhi-update/watch-bodhi-requests.py
@@ -197,7 +197,7 @@ when new requests are filed in bodhi')
         try:
             new_updates = bodhi_new_requests_since(r, int(lastcheck), testing_cachefile,
                                                     updatecache=(not opts.dryrun))
-        except fedora.client.FedoraServiceError as e:
+        except fedora.client.FedoraServiceError, e:
             print "ERROR: %s" % e
         # Sort the new updates based on their target/request
         updates = {'testing': [], 'stable': [], 'critpath': []}
@@ -213,7 +213,7 @@ when new requests are filed in bodhi')
             new_updates = []
             new_updates = bodhi_new_stable_requests(r, stable_cachefile,
                                                updatecache=(not opts.dryrun))
-        except fedora.client.FedoraServiceError as e:
+        except fedora.client.FedoraServiceError, e:
             print "ERROR: %s" % e
         print "%i stable update requests found" % len(new_updates)
         updates['stable'] += new_updates

[PATCH 3/3] Correct call to util.get_basearch()

[James Laska]

---
 tests/rats_sanity/rats_sanity.py |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/tests/rats_sanity/rats_sanity.py b/tests/rats_sanity/rats_sanity.py
index 1da8e32..8ae7081 100644
--- a/tests/rats_sanity/rats_sanity.py
+++ b/tests/rats_sanity/rats_sanity.py
@@ -39,7 +39,7 @@ class rats_sanity(AutoQATest):
     @ExceptionCatcher("self.run_once_failed")
     def run_once(self, baseurl, parents, name, **kwargs):
         if name:
-            name = "%s-%s" % (name, autoqa.util.get_basearch())
+            name = "%s-%s" % (name, util.get_basearch())
         else:
             name = baseurl
         os.chdir(self.bindir)

[PATCH 3/3] Correct call to util.get_basearch()

[Will Woods]

On Tue, 2010-09-07 at 16:13 -0400, James Laska wrote:
> tests/rats_sanity/rats_sanity.py |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)

These 3 patches all look fine. I'd say "go ahead and apply to master"
but I'm pretty sure you're targeting these things for 0.4.0 and we're
already a little ways toward 0.4.1.

Since Kamil was thinking about reworking the multihook_tests patch,
maybe that merge should get reverted so you can just apply these small
patches to master, and package master as 0.4.0.

What do you think?

-w

[PATCH 3/3] Correct call to util.get_basearch()

[James Laska]

On Tue, 2010-09-07 at 17:08 -0400, Will Woods wrote:
> On Tue, 2010-09-07 at 16:13 -0400, James Laska wrote:
> > tests/rats_sanity/rats_sanity.py |    2 +-
> >  1 files changed, 1 insertions(+), 1 deletions(-)
>
> These 3 patches all look fine. I'd say "go ahead and apply to master"
> but I'm pretty sure you're targeting these things for 0.4.0 and we're
> already a little ways toward 0.4.1.
>
> Since Kamil was thinking about reworking the multihook_tests patch,
> maybe that merge should get reverted so you can just apply these small
> patches to master, and package master as 0.4.0.
>
> What do you think?

No objections here.  Some additional data ... there are a few other
issues lurking which I've not managed to triage/diagnose yet.  They
include ...

1. rpmguard test is sending emails even when no previous package is
        found -
        https://fedorahosted.org/pipermail/autoqa-results/2010-September/032313.html
     2. rpmguard run_once() crash -
        http://autoqa.fedoraproject.org/results/581-autotest/qa04.c.fedoraproject.or...
     3. rats_sanity crash (should be fixed by previous patch) -
        http://autoqa.fedoraproject.org/results/570-autotest/qa04.c.fedoraproject.or...
     4. conflicts crashes -
        http://autoqa.fedoraproject.org/results/568-autotest/qa04.c.fedoraproject.or...
     5. repoclosure crashes -
        http://autoqa.fedoraproject.org/results/554-autotest/qa04.c.fedoraproject.or...

I'll try to pinpoint these this week and send patches for review.  Once
those are knocked out, maybe we can see where the multihook patchset
is ... and adjust accordingly?

Thanks,
James

[PATCH 3/3] Correct call to util.get_basearch()

[Kamil Paral]

----- "James Laska"  wrote:
> No objections here.  Some additional data ... there are a few other
> issues lurking which I've not managed to triage/diagnose yet.  They
> include ...
>  1. rpmguard test is sending emails even when no previous package
>
> is
>         found -
>
> https://fedorahosted.org/pipermail/autoqa-results/2010-September/032313.html

That is correct. Since AutoQATest parent class now handles email sending,
we send out test result every time. We can argue whether we want to have some
option to disable sending emails in certain cases, but I would say it's just
fine this way. Certainly in this case - the test passed and we want to know it. 
Just summary and highlights are mixed up, I'll fix that in my forthcoming 
patchset.
>  2. rpmguard run_once() crash -
>
> http://autoqa.fedoraproject.org/results/581-autotest/qa04.c.fedoraproject.or...

A typo (not mine:)). I'll fix that in my patchset.
>  3. rats_sanity crash (should be fixed by previous patch) -
>
> http://autoqa.fedoraproject.org/results/570-autotest/qa04.c.fedoraproject.or...

Yes, your patch should fix this.
>  4. conflicts crashes -
>
> http://autoqa.fedoraproject.org/results/568-autotest/qa04.c.fedoraproject.or...

Not sure what this is, maybe just network failure?
>  5. repoclosure crashes -
>
> http://autoqa.fedoraproject.org/results/554-autotest/qa04.c.fedoraproject.or...

That's this one - https://fedorahosted.org/autoqa/ticket/214. I wonder what
we can do about it to fix this faster. It may take a long time until
new upstream yum-utils are releases and they get into Fedora and EPEL.
Do we want to create our own devel build from VCS?
> I'll try to pinpoint these this week and send patches for review. 
> Once
> those are knocked out, maybe we can see where the multihook patchset
> is ... and adjust accordingly?

Don't bother with those rpmguard issues, I'll handle them. Thanks.

[PATCH 3/3] Correct call to util.get_basearch()

[James Laska]

On Thu, 2010-09-09 at 03:15 -0400, Kamil Paral wrote:
> ----- "James Laska"  wrote:
> > No objections here.  Some additional data ... there are a few other
> > issues lurking which I've not managed to triage/diagnose yet.  They
> > include ...
> >  1. rpmguard test is sending emails even when no previous package
> >
> > is
> >         found -
> >
> > https://fedorahosted.org/pipermail/autoqa-results/2010-September/032313.html
>
> That is correct. Since AutoQATest parent class now handles email sending,
> we send out test result every time. We can argue whether we want to have some
> option to disable sending emails in certain cases, but I would say it's just
> fine this way. Certainly in this case - the test passed and we want to know it. 
> Just summary and highlights are mixed up, I'll fix that in my forthcoming 
> patchset.

Ah okay ... that's the issue, that the subject and body of the message
are mixed.  Something seemed weird and I couldn't pinpoint it :)
> >  2. rpmguard run_once() crash -
> >
> > http://autoqa.fedoraproject.org/results/581-autotest/qa04.c.fedoraproject.or...
>
> A typo (not mine:)). I'll fix that in my patchset.

Thanks :)
> >  5. repoclosure crashes -
> >
> > http://autoqa.fedoraproject.org/results/554-autotest/qa04.c.fedoraproject.or...
>
> That's this one - https://fedorahosted.org/autoqa/ticket/214. I wonder what
> we can do about it to fix this faster. It may take a long time until
> new upstream yum-utils are releases and they get into Fedora and EPEL.
> Do we want to create our own devel build from VCS?

I spoke to Seth on #fedora-devel regarding plans for a newer yum-utils
build.  He isn't anticipating one soon, and recommended doing a
custom-build that includes the referenced upstream patch.  I've filed a
ticket to remind me to take care of this ...

https://fedorahosted.org/autoqa/ticket/225
> > I'll try to pinpoint these this week and send patches for review. 
> > Once
> > those are knocked out, maybe we can see where the multihook patchset
> > is ... and adjust accordingly?
>
> Don't bother with those rpmguard issues, I'll handle them. Thanks.

Okay, thanks Kamil!

Thanks,
James